PSParseHTML.psm1
|
Function ConvertFrom-HtmlTable { [cmdletbinding()] param ([Parameter(Mandatory = $true)] [string]$Content) Begin { $HTMLParser = [AngleSharp.Html.Parser.HtmlParser]::new() } Process { $ParsedDocument = $HTMLParser.ParseDocument($content) [Array] $Tables = $ParsedDocument.GetElementsByTagName('table') :table foreach ($table in $tables) { $headers = $table.Rows[0].Cells.TextContent.Trim() | Where-Object { $_ } if ($null -ne $headers) { [Array] $output = foreach ($row in $table.Rows | Select-Object -Skip 0) { if (@($row.Cells).count -ne $headers.count) { Write-Verbose 'Unsupported table.' Continue table } $obj = [ordered]@{ } for ($x = 0; $x -lt $headers.count; $x++) { $obj["$($headers[$x])"] = $row.Cells[$x].TextContent.Trim() } [PSCustomObject] $obj } if ($output.count -ge 1) { @(, $output) } else { Write-Verbose 'Table has no rows' } } } } End { } } function Convert-HTMLToText { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Convert-HTMLToText - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Convert-HTMLToText - No choice file or Content. Termninated.' return } $Output = Convert-InternalHTMLToText -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Convert-InternalHTMLToText { [CmdletBinding()] param([string] $Content) $Output = [NUglify.Uglify]::HtmlToText($Content) if ($Output.HasErrors) { Write-Warning "Convert-HTMLToText -Errors: $($Output.Errors)" } $Output.Code } function Format-CSS { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Format-CSS - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Format-CSS - No choice file or Content. Termninated.' return } $Output = Format-InternalCSS -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Format-HTML { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Format-HTML - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Format-HTML - No choice file or Content. Termninated.' return } $Output = Format-InternalHTML -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Format-InternalCSS { [CmdletBinding()] param([string] $Content) $CssParser = [AngleSharp.Css.Parser.CssParser]::new() $ParsedDocument = $CssParser.ParseStyleSheet($Content) $StringWriter = [System.IO.StringWriter]::new() $PrettyMarkupFormatter = [AngleSharp.Css.CssStyleFormatter]::new() $ParsedDocument.ToCss($StringWriter, $PrettyMarkupFormatter) $StringWriter.ToString() } function Format-InternalHTML { [CmdletBinding()] param([string] $Content) $HTMLParser = [AngleSharp.Html.Parser.HtmlParser]::new() $ParsedDocument = $HTMLParser.ParseDocument($Content) $StringWriter = [System.IO.StringWriter]::new() $PrettyMarkupFormatter = [AngleSharp.Html.PrettyMarkupFormatter]::new() $ParsedDocument.ToHtml($StringWriter, $PrettyMarkupFormatter) $StringWriter.ToString() } function Format-InternalJS { [CmdletBinding()] param([string] $Content, [int] $IndentSize = 4, [string] $IndentChar = ' ', [bool] $IndentWithTabs = $false, [bool] $PreserveNewlines = $true, [double] $MaxPreserveNewlines = 10.0, [bool] $JslintHappy = $false, [Jsbeautifier.BraceStyle] $BraceStyle = [Jsbeautifier.BraceStyle]::Collapse, [bool] $KeepArrayIndentation = $false, [bool] $KeepFunctionIndentation = $false, [bool] $EvalCode = $false, [int] $WrapLineLength = 0, [bool] $BreakChainedMethods = $false) $Jsbeautifier = [Jsbeautifier.Beautifier]::new() $Jsbeautifier.Opts.IndentSize = $IndentSize $Jsbeautifier.Opts.IndentChar = $IndentChar $Jsbeautifier.Opts.IndentWithTabs = $IndentWithTabs $Jsbeautifier.Opts.PreserveNewlines = $PreserveNewlines $Jsbeautifier.Opts.MaxPreserveNewlines = $MaxPreserveNewlines $Jsbeautifier.Opts.JslintHappy = $JslintHappy $Jsbeautifier.Opts.BraceStyle = $BraceStyle $Jsbeautifier.Opts.KeepArrayIndentation = $KeepArrayIndentation $Jsbeautifier.Opts.KeepFunctionIndentation = $KeepFunctionIndentation $Jsbeautifier.Opts.EvalCode = $EvalCode $Jsbeautifier.Opts.WrapLineLength = $WrapLineLength $Jsbeautifier.Opts.BreakChainedMethods = $BreakChainedMethods $FormattedJS = $Jsbeautifier.Beautify($Content) $FormattedJS } function Format-JavaScript { [alias('Format-JS')] [CmdletBinding()] param([string] $File, [string] $OutputFile, [alias('FileContent')][string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Format-JavaScript - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Format-JavaScript - No choice file or Content. Termninated.' return } [int] $IndentSize = 4 [string] $IndentChar = ' ' [bool] $IndentWithTabs = $false [bool] $PreserveNewlines = $true [double] $MaxPreserveNewlines = 10.0 [bool] $JslintHappy = $false [Jsbeautifier.BraceStyle] $BraceStyle = [Jsbeautifier.BraceStyle]::Collapse [bool] $KeepArrayIndentation = $false [bool] $KeepFunctionIndentation = $false [bool] $EvalCode = $false [int] $WrapLineLength = 0 [bool] $BreakChainedMethods = $false $SplatJS = @{IndentSize = $IndentSize IndentChar = $IndentChar IndentWithTabs = $IndentWithTabs PreserveNewlines = $PreserveNewlines MaxPreserveNewlines = $MaxPreserveNewlines JslintHappy = $JslintHappy BraceStyle = $BraceStyle KeepArrayIndentation = $KeepArrayIndentation KeepFunctionIndentation = $KeepFunctionIndentation EvalCode = $EvalCode WrapLineLength = $WrapLineLength BreakChainedMethods = $BreakChainedMethods } $Output = Format-InternalJS -Content $Content @SplatJS if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Optimize-CSS { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Optimize-CSS - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Optimize-CSS - No choice file or Content. Termninated.' return } $Output = Optimize-InternalCSS -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Optimize-HTML { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Optimize-HTML - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Optimize-HTML - No choice file or Content. Termninated.' return } [bool] $ShouldKeepAttributeQuotes = $true [bool] $ShouldKeepComments = $true [bool] $ShouldKeepEmptyAttributes = $true [bool] $ShouldKeepImpliedEndTag = $true [bool] $ShouldKeepStandardElements = $true $Output = Optimize-InternalUglifyHTML -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } function Optimize-InternalCSS { [CmdletBinding()] param([string] $Content) $CSSParser = [AngleSharp.Css.Parser.CssParser]::new() $ParsedDocument = $CSSParser.ParseStyleSheet($Content) $StringWriter = [System.IO.StringWriter]::new() $PrettyMarkupFormatter = [AngleSharp.Css.MinifyStyleFormatter]::new() $ParsedDocument.ToCss($StringWriter, $PrettyMarkupFormatter) $StringWriter.ToString() } function Optimize-InternalUglifyCSS { [CmdletBinding()] param([string] $Content) [NUglify.Uglify]::Css($Content).Code } function Optimize-InternalUglifyHTML { [CmdletBinding()] param([string] $Content) $Settings = [NUglify.Html.HtmlSettings]::new() $Settings.RemoveOptionalTags = $false [NUglify.Uglify]::Html($HTMLContentFormatted, $Settings).Code } function Optimize-InternalUglifyJS { [CmdletBinding()] param([string] $Content) [NUglify.Uglify]::Js($Content).Code } function Optimize-JavaScript { [CmdletBinding()] param([string] $File, [string] $OutputFile, [string] $Content) if ($File) { if (Test-Path -LiteralPath $File) { $Content = [IO.File]::ReadAllText($File) } else { Write-Warning "Optimize-JavaScript - File doesn't exists" return } } elseif ($Content) { } else { Write-Warning 'Optimize-JavaScript - No choice file or Content. Termninated.' return } $Output = Optimize-InternalUglifyJS -Content $Content if ($OutputFile) { [IO.File]::WriteAllText($OutputFile, $Output) } else { $Output } } if ($PSEdition -eq 'Core') { Add-Type -Path $PSScriptRoot\Lib\Core\AngleSharp.Css.dll Add-Type -Path $PSScriptRoot\Lib\Core\AngleSharp.dll Add-Type -Path $PSScriptRoot\Lib\Core\jint.dll Add-Type -Path $PSScriptRoot\Lib\Core\Jsbeautifier.dll Add-Type -Path $PSScriptRoot\Lib\Core\NUglify.dll } else { Add-Type -Path $PSScriptRoot\Lib\Default\AngleSharp.Css.dll Add-Type -Path $PSScriptRoot\Lib\Default\AngleSharp.dll Add-Type -Path $PSScriptRoot\Lib\Default\jint.dll Add-Type -Path $PSScriptRoot\Lib\Default\Jsbeautifier.dll Add-Type -Path $PSScriptRoot\Lib\Default\NUglify.dll Add-Type -Path $PSScriptRoot\Lib\Default\System.Text.Encoding.CodePages.dll } Export-ModuleMember -Function @('ConvertFrom-HtmlTable', 'Convert-HTMLToText', 'Format-CSS', 'Format-HTML', 'Format-JavaScript', 'Optimize-CSS', 'Optimize-HTML', 'Optimize-JavaScript') -Alias @('Format-JS') |